BlenSor: Blender Sensor Simulation Toolbox

By Michael Gschwandtner, Roland Kwitt, Andreas Uhl, Wolfgang Pree. 2011

301 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Networks and Communications, Instrumentation, 3D Vision & Pose, Robotics & Control

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

1. Sergey I. Nikolenko
2. Binh Yang
3. Andreas Geiger
4. Raquel Urtasun
5. Shenlong Wang
6. Vijay Badrinarayanan
7. Roberto Cipolla
8. Michael Firman

In Computer Networks and Communications

1. Charles Ruizhongtai Qi
2. Niloy J. Mitra
3. Sergey I. Nikolenko
4. Stefan Schaal
5. Mehmet Ersin Yümer
6. Raquel Urtasun
7. Ankur Handa
8. Binh Yang

In Instrumentation

1. Charles Ruizhongtai Qi
2. Niloy J. Mitra
3. Sergey I. Nikolenko
4. Roberto Cipolla
5. Mehmet Ersin Yümer
6. Raquel Urtasun
7. Binh Yang

In 3D Vision & Pose

1. Sergey I. Nikolenko
2. Raquel Urtasun
3. M. Ovsjanikov
4. Charles Ruizhongtai Qi
5. Binh Yang
6. Michael Firman
7. Simon Stent
8. Roberto Cipolla

In Robotics & Control

1. Sergey I. Nikolenko
2. Michael Firman
3. Daniel Kappler
4. Stefan Schaal
5. Charles Ruizhongtai Qi
6. Sergey Zakharov
7. Niloy J. Mitra
8. Roberto Cipolla

Citations in the corpus

• RGBD Datasets: Past, Present and Future (2016, background): “While sensor noise can be emulated [44, 40, 29], it can be very difficult for synthetic scenes to capture the true properties of the real world.”
• DepthSynth: Real-Time Realistic Synthetic Data Generation from CAD Models for 2.5D Recognition (2017, methodology): “[25, 24], from BlenSor [17] and from 3 Kinect error models—respectively from Menna et al.”
• Understanding RealWorld Indoor Scenes with Synthetic Data (2015, methodology): “We realise the possible mismatch of the distribution of the noise characteristics in real world datasets and our synthetic depth-maps and therefore add noise to the perfect rendered depth-maps according to the simulated kinect noise model in [14, 20] (see Fig.”
• Robot arm pose estimation by pixel-wise regression of joint angles (2016, methodology): “To make the images as realistic as possible, we used a reimplementation of the sensor model proposed in [31], which simulates typical effects that appear in images of real RGB-D cameras like the Kinect or Xtion (e.”
• Synthetic Data for Deep Learning (2019, methodology): “It simulates Mobile Laser Scanner (MLS) readings with a Blender plugin [214] and is specifically intended for pretraining deep neural networks.”
• PointCleanNet: Learning to Denoise and Remove Outliers from Dense Point Clouds (2019, methodology): “We evaluated our method on a simulated LIDAR dataset ( Velo-dyne ) generated using BlenSor [GKUP11], which models various types of range scanners.”
• UniSim: A Neural Closed-Loop Sensor Simulator (2023, unknown): “Most prior sensor simulation systems use 3D-scanned or manually built synthetic environments for small indoor environments [33, 37, 67], and perform rasterization or ray-tracing [58,69] to simulate various sensor data [20,22,30].”
• Learning Semantic Deformation Flows with 3D Convolutional Networks (2016, methodology): “We train and test on three different data types: 3D shapes in mesh representation, point sets sampled directly on the shape, and simulated single frame depth scans from a kinect sensor [15] in point cloud representation.”
• LiDARsim: Realistic LiDAR Simulation by Leveraging the Real World (2020, methodology): “Based on these graphics engines, simulators have been developed to provide virtual sensor simulation such as CARLA and Blensor [8, 14, 39, 16, 47].”
• Deep Learning on Point Sets for 3 D Classification and Segmentation (2016, methodology): “For every CAD model in the ShapeNet part data set, we use Blensor Kinect Simulator [7] to generate incomplete point clouds from six random viewpoints.”